High spec EJ257 misfire, cam correlation, cranking compression difference between two banks

[Rozsko]

Hey everyone,

This will be a bit long, I am sorry about that, but I think knowing the history might help to understand the situation better.

So, given a 07 EDM Hawk STi which I bought 4 years ago and had some serious coolant leaking problem with it. long story short, I ended up replacing the whole engine with an IAG closed deck block, forged internals, Cosworth heads with matching cams, Litchfield LM450 twinscroll turbo, ID 1050X injectors (had DW850 before that), etc...

After the rebuild I was planning to tune the car myself but when I was trying to do MAF calibration (due to APS CAI being installed), the signal and the AFR readings were erratic, and Idle was sort of rough, so before moving on to the actual tune I wanted figure out if this is something wrong or just the nature of all the mods on the engine. Initially I did not realize but there were some misfire as well. Not enough to throw CEL, but definitely higher then normal (NO lightweight flywheel or crank sprocket).

I bought a scope and started to look into sensor readings. While the signals were generally noisy from the ignition, everything seemed to be fine, no matter what I "fixed", no actual improvement was achieved.

The only (I guess real) problem I found was the difference between left and right bank cranking compression test done with an amp clamp on the battery. There is about 15% difference in amperage and I would consider that not normal. I tried to do the a running compression test as well, but the transducer I built is far from superior, so that I was not able to use to diagnose anything.

An other (real or believed) issue I found is a 5 crank degree misalignment of the left and right cam signal (this I suspected is due to timing belt jump, but that would be 15 degrees and I checked the markings and all wheel markings are lining up perfectly). Additionally I had something similar to the symptoms found in here: 

which is the signal of the LH cam sensor being inverting back and forth randomly at idle. Based on some technical reading this is usual with these hall sensors at low engine speed, but rotating the sensor with 180 degrees solved the issue for some and for me as well. Now that on it's own seems to effect the sensor calibration and I had the set the offset to 104 degrees instead of the OE value of 126. The cam solenoids seem to work fine, as I tested them by stepping through the vales from 0 to 40 degrees at around 1500RMP and the engine was almost stalling at 40 degrees and the cam sensor readings followed perfectly the commanded solenoid values.

To be honest even today I am not sure if the square wave signals should both go from 5V to 0V, or one side should go 5 to 0 and the other from 0 to 5.

Well that was little jump forward in time, as in the meanwhile I was able to source a secondhand WRXLINK107 and since then I am playing with that.

The engine is running fine, I mean idles and runs well (without boost control), the AFR readings are more stable (much more) with the SD model, but I still have the misfire at idle and this along with the the left/right compression test difference and the cam signal misalignment still concerns me and I am really not sure if I could start the power tuning or should I do some more mechanical tests (which most probably would mean I need to pull the engine), but I threw too much money into the rebuild to scrap it during the tuning process (though I also bought a knockblock).

Any hints, suggestions would be appreciated.

Thanks,

Béla

current map + AFR target.pclr

[Adamw]

Since there is some evidence of a compression issue based on your starter current, the first thing I would do is buy a cheap compression gauge and do a basic warm WOT compression test.  Get that possibility eliminated first. 

[Rozsko]

9 hours ago, Adamw said:

Since there is some evidence of a compression issue based on your starter current, the first thing I would do is buy a cheap compression gauge and do a basic warm WOT compression test.  Get that possibility eliminated first. 

Yes, that is something on my list of to-dos. That is just a freaking pain in the back, being a flat 4 engine.

But is it likely that two cylinders on the same side having exactly the same compression issue due to internal mechanical problem? This is why I was investigating from a cam side as that would equally effect two cylinders at the same time.

Anyhow, next week I will do the compression test.

[Adamw]

11 hours ago, Rozsko said:

But is it likely that two cylinders on the same side having exactly the same compression issue due to internal mechanical problem?

Since it is an adjacent pair of cylinders there are many possibilities.  I have seen all sorts of fails over the years, so you cant assume anything, you need to eliminate the most obvious/logical possibilities as completely as you can first then move on to the next.  Just some random examples of what I come across before that could potentially affect compression on 1 bank;  Cyl head bolts/studs bottomed out on the threads before the head clamps tight. Head gaskets on upside down or back to front.  Small fire ring gasket on oversize bore causing leak between cylinders.  Valve lash too tight or lifters not bled. Cam timing.  I have even come across things like rags left inside the ports/manifold... 

[Stevieturbo]

If you have a scope, you should be able to do a running compression test quite easily. I've done it before using a cheap 300psi absolute pressure sensor off ebay.

 

At higher rpm's there is a little lag with the sensor, but for the purposes of this test it is viable. I just bought a cheap compression tester and adapted the sensor to the hose so it can be used as normal. A little awkward in a Subaru maybe, but still very doable.

 

So this new engine itself....has never actually ran correctly ? What does a 4 gas analyser tell you ? Is there evidence of misfires ?

[Rozsko]

20 minutes ago, Stevieturbo said:

If you have a scope, you should be able to do a running compression test quite easily. I've done it before using a cheap 300psi absolute pressure sensor off ebay.

 

At higher rpm's there is a little lag with the sensor, but for the purposes of this test it is viable. I just bought a cheap compression tester and adapted the sensor to the hose so it can be used as normal. A little awkward in a Subaru maybe, but still very doable.

 

So this new engine itself....has never actually ran correctly ? What does a 4 gas analyser tell you ? Is there evidence of misfires ?

thanks a lot for reply.

I already made a similar compression tester, but I did not trust it as the shape of the curve on the scope is too squary (if that is a word) as I assume the response time of the eBay sensor is too high.

As far as 4 gas analyzer, I never tried that as I don't have one, but the misfire is pretty audible when it misses the beat.

[Stevieturbo]

This is a screenshot from a datalog on my Syvecs. Have removed most info from it for clarity, but upper graph is crank/cam ( 36-1 and std LS 1x cam ) and lower cyl pressure in cylinder 1 from initial cranking to actually running/idling.

Upper graph also shows battery voltage. Most are logged at 1000Hz for the purposes of the test. I wouldnt say the pressure sensor is calibrated 100% correctly but it should be pretty close. I'm sure it was a 300psi absolute sensor. Although I had bought 300 and 500psi sensors so cant remember which I used here. Not that it should matter. I just found that assuming a calibration based on 0.5 to 4.5v for the range it should have been...just seemed to give wrong numbers. And my Mityvac only went to 60psi positive. I have a small reg for my airline to go higher, but damned if I could find it when I did this.

Oddly the first couple of bumps read very high pressure, compared to once it actually gets moving. As this was a start and run test, obviously it was not done with a fully open throttle as a normal compression test with a non running engine would be

I presume you've tried the obvious of new plugs ?

Coil problems are rare, but some do fail. Can you swap the coils ?

Common cause of misfires on standard engines is valve/seat recession...but if you're using new heads that have been built/shimmed correctly etc...that should not be an issue.

[Rozsko]

On 9/1/2019 at 2:11 AM, Stevieturbo said:

This is a screenshot from a datalog on my Syvecs. Have removed most info from it for clarity, but upper graph is crank/cam ( 36-1 and std LS 1x cam ) and lower cyl pressure in cylinder 1 from initial cranking to actually running/idling.

Upper graph also shows battery voltage. Most are logged at 1000Hz for the purposes of the test. I wouldnt say the pressure sensor is calibrated 100% correctly but it should be pretty close. I'm sure it was a 300psi absolute sensor. Although I had bought 300 and 500psi sensors so cant remember which I used here. Not that it should matter. I just found that assuming a calibration based on 0.5 to 4.5v for the range it should have been...just seemed to give wrong numbers. And my Mityvac only went to 60psi positive. I have a small reg for my airline to go higher, but damned if I could find it when I did this.

Oddly the first couple of bumps read very high pressure, compared to once it actually gets moving. As this was a start and run test, obviously it was not done with a fully open throttle as a normal compression test with a non running engine would be

I presume you've tried the obvious of new plugs ?

Coil problems are rare, but some do fail. Can you swap the coils ?

Common cause of misfires on standard engines is valve/seat recession...but if you're using new heads that have been built/shimmed correctly etc...that should not be an issue.

Thanks Stevie for your reply again.

First of all, how are you measuring the in cylinder pressure with an ECU? You wired it up just for the sake of it?

Unfortunately I had no time to do the manual compression test before I went back to Italy (as I am working there at the moment and I go home only ever other month), but I found the scope trace saves on my laptop when I was doing the compression test on cyl#2 and 3,and as you can see the test shows an 11.5 % difference between the two banks. Again, I am not sure if the slow response of the 300psi sensor has an effect on the measurement or not.

On the spark plugs, yes. I replaced those although the ones installed were brand new Denso plugs with 1 heat range colder. The 1 range colder thing aplies to the replacement plugs too.

Regarding to the coils, I swapped them and also replaced one of them as I smoked that 4 times due to leaving the scope unplugged but connected to the spark plug connector.

I have no idea about the valves. When the heads were put back together, we checked the lash and they were good, but whether they are seated correctly or not, I don't know. The valves are oversized, so there were some obvious machine work involved.

[Steve]

Use the rotional rulers to check cam events like below. Compare sides.

 

 

[Stevieturbo]

I want a Pico !

 

But yes, I just rigged up the pressure sensor to use the ecu to record a cranking and running compression test....because why not ? lol I have an ecu that can....and I do not have a Pico that can. Pico would be a lot more useful, although even with your 11% discrepancy bank to bank, I dont honestly think that would be enough to start pointing to a problem.

What plugs, what gaps ? What were your valve clearances ? A little larger is usually a good thing, OEM specs are pretty tight.

And was valve sealing checked prior to rebuilding ? You say Cosworth heads...so does that mean they were supplied complete and ready to go, or where multiple parties involved in putting them together ?

[Rozsko]

 

4 hours ago, Stevieturbo said:

I just rigged up the pressure sensor to use the ecu to record a cranking and running compression test.

That's an awesome idea. I would never have thought about doing that.

4 hours ago, Stevieturbo said:

What plugs, what gaps ? What were your valve clearances ? A little larger is usually a good thing, OEM specs are pretty tight.

Plugs I tried both Denso and NGK, 1 range colder then OEM, but don't remember the code of those.

Valve clearance (if you mean valve lash) was checked when we bolted on the the heads, but again can't remember. The tech who did the assembly said they are all good.

4 hours ago, Stevieturbo said:

And was valve sealing checked prior to rebuilding ? You say Cosworth heads...so does that mean they were supplied complete and ready to go, or where multiple parties involved in putting them together ?

Valve sealing were never checked, or at least I am not aware if were.

Heads are Cosworth CNC ported, but were not coming as 1 assembled unit. Valves (oversized Supertech) were machined and fitted into the heads separately and the Cosworth cams were fitted after that. So different parteis were involved, yes.

This week I had not much time to play with the pressure graphs, but next week I will take a deeper look at them.

The other thing that bothers me is the shape of the secondary ignition wave form. They are nothing like the ones I would know to be good.

oh, and thanks for the tip on that Windows app. Again, I would have never thought that you can find anything like that in the Windows app store.

[Adamw]

On 9/3/2019 at 10:38 AM, Rozsko said:

I was doing the compression test on cyl#2 and 3,and as you can see the test shows an 11.5 % difference between the two banks.

Most pressure sensors output 0.5V for zero pressure, so if you remove that offset, your difference is more like 20%.

 

On 9/3/2019 at 10:38 AM, Rozsko said:

I am not sure if the slow response of the 300psi sensor has an effect on the measurement or not.

Are you sure it was a 300psi sensor?  If its a 300psi sensor and you only have a max of 1.2V signal, that suggest your compression is only about 70psi.

[Rozsko]

On 9/9/2019 at 11:46 AM, Adamw said:

Are you sure it was a 300psi sensor?

That is what I ordered for sure, but I will ask my wife at home to check the text on it.

 

On 9/9/2019 at 11:46 AM, Adamw said:

your difference is more like 20%

Good point and I can't wait to get home again mid oct to finally do a static compression test.

[Stevieturbo]

Ideally it would be a 300psi absolute sensor, so it can also read vacuum.

 

Cant comment on the ignition waveform though, just not something I've had enough experience with. Some great videos on youtube from Scannerdanner though, and no doubt assistance via Pico forum etc.

[Rozsko]

13 hours ago, Stevieturbo said:

Some great videos on youtube from Scannerdanner though

yea, I am trough those (and even more) and have purchased the ebook from him as well, but if you look at the waveform above, those are like white and black which leads me to think one of two things: 1) I am doing something insanely wrong when recording sec. ign signal, 2) there are some very bad things going on inside my engine's cylinders.

[Stevieturbo]

How are you actually measuring the sec waveform ? With a lead or something stuffed into the coil ? Or via some other adapter ?

 

IMO if the coils are known good, plugs are known good, and wiring etc is known good...I think this area is of little concern. OEM coils are very powerful and reliable. So unless you're doing something strange in terms of dwell times or other....the problem is elsewhere.

I guess a current clamp on the supply to all 4 coils might show any discrepancy between them ? Probably a waste of time though

I've seen a lot of standard type cars with valve sealing issues due to valve/seat erosion, more so earlier 2000 model cars, say 2001-2006 though. So saying valve lash is "good".....actual numbers would matter more than "good".

 

I did read years ago, that any issues like this, first port of call should be injectors. I can see their reasoning behind the statement, although in reality it isnt the first port of call. But I dont think you've mentioned what you're using, or how they're setup or configured within the ecu ?

 

 

[Rozsko]

8 hours ago, Stevieturbo said:

How are you actually measuring the sec waveform ?

I have an ignition lead going from the coil to the plug and have a Hantek capacitive clamp attached to it.

8 hours ago, Stevieturbo said:

I guess a current clamp on the supply to all 4 coils might show any discrepancy between them ?

Never thought to measure current on them, but I can try. I spend so much time on this yet, so a little more doesn't matter.

8 hours ago, Stevieturbo said:

actual numbers would matter more than "good"

I know. Unfortunately at the time the block was put together, I wasn't focusing on these things, as I have never thought it will go wrong. And because of that I did not record these numbers and was totally relying on the tech who put together the engine.

8 hours ago, Stevieturbo said:

But I dont think you've mentioned what you're using, or how they're setup or configured within the ecu ?

Ok, so after the build, I had brand new DW850 incetors, and recently I changed to a set of ID1050X as they provide PnP calibration data for the Link ECUs. These are brand new too.

 

[Steve]

Here is a treat for you. It`ll be worth it, i promise.

http://www.tstseminars.org/webinars/premium-webinars/tst-january-24th-2019-simulcast-pressure-acquisition-analysis-from-the-inside-out-w-brandon-steckler?fbclid=IwAR31ABBbG1EMAbQH3o_XCdzwpLLKLbAv0kNJLCMf3cS28JsJlOYKSVtfbxI

[Rozsko]

38 minutes ago, Steve said:

Here is a treat for you. It`ll be worth it, i promise.

http://www.tstseminars.org/webinars/premium-webinars/tst-january-24th-2019-simulcast-pressure-acquisition-analysis-from-the-inside-out-w-brandon-steckler?fbclid=IwAR31ABBbG1EMAbQH3o_XCdzwpLLKLbAv0kNJLCMf3cS28JsJlOYKSVtfbxI

Ha ha!

I love tst seminars.

I already have payed for couple of their premium videos.

This one will be a nice addition to the list. Thanks

[Stevieturbo]

SD linked to this ( first of three though ) a few days ago.

 

Free and nearly an hour long. I downloaded it to watch, although havent got around to it yet. Of course biggest aspect of all this. Is having the correct type of transducer and more importantly software to go with it.

 

Another I linked to before which was very interesting. This guy sells a scope setup from Ukraine and has done a lot of videos. With some very interesting software to go along with them. Obviously not applicable to all engines, but interesting all the same.

 

https://www.youtube.com/channel/UCZfEJfHc8DpV6G23jClwruA/videos

 

eg 

 

[Rozsko]

Little update, jut so that you have something to read.

I contacted the USB Autoscope guys in Ukraine. The whole package is not too expensive (well, relative to the Pico stuff), but the only way they accept payment is through Western Union. Unfortunately they don't offer any buyer protection, like PayPal, so I was considering driving the car to Ukraine, as it is neighborhood country to Hungary. The headquarter is quite far, but they recommended some techs near the border who can do the diagnostic. This way I could avoid buying the kit and pay only for the diagnostic. So I was ready to roll, but then they figured out my car is far from stock and they backed out saying it would require a lot more time to diagnose as the standard process is for OE setups.

Long story short, I ended up ordering a Pico WPS transducer, although pulling the engine and disassemble it would be cheaper, I just don't prefer that until I am really convinced it is necessary.

Anyhow, I am not sure if the transducer will arrive next week, while I am home. If it will, then I will post up some more pics on the results, if not then couple of more weeks will need to pass, until I get my hands dirty again.

[Stevieturbo]

Yea, the Western Union put me off too. If they accepted Paypal, I'd have bought their kit.

 

I would still be pulling the valve covers off to check valve clearances though. It's awkward to do, but at least you'd know for sure if they are ok or not.

 

Not sure how tight clearances would show on a running test as never done one, but I have had cars before where that is the problem, but neither compression tests or leakdown tests revealed anything unusual.

 

[Steve]

Good choice on the WPS. I know it hurts the valet like a mofo (as i got one myself and really want another one in addition), but its THE pressuretransducer to get at this day.

You can clearly see valve events and put up rotational rulers to measure at what crank degree they happen. Then you look at the odd man out. IE you compare to the other cylinders. A to tight valve clearence would make the valve open early and close late. There is plenty situations that a conventional compression test or leakdowntest wont show.

[Stevieturbo]

When I used a cheapo sensor, there is noticable lag from pressure peaks ( ie TDC ) in relation to my cam trigger as rpm rises. I assume this is simply sensor lag ?

The WPS quotes 100us...which is really 1ms. Which in the grand scheme of things, isnt that fast really ? Their sensor is a hell of a lot of money

I think the cheapo sensor I used claimed 2ms response ( indeed searching most brands, they just claim <2ms anyway )

 

I think some of the brand name sensors I seen were saying around 1ms too, like Honeywell etc. I just bought the cheap one first as a test. I guess the Autoscope one would be decent enough too.

[Rozsko]

7 hours ago, Stevieturbo said:

100us...which is really 1ms.

100us=0.1ms

But other then that I fully agree. I bought a cheap one from eBay to see how it is going and it proves to be wrong.

Now my worry is the scope itself, as I did not go with the Pico autodiagnostic one, but the generic only which is 8bit vertical resolution.

The only problem is that I won't have the wps before I need to leave home next Monday. So it will be an other few weeks until I can get to the car again.